Quantitative Analysis of the game of Survivor...of course!
First and foremost, welcome to the site and please take some time to explore! This site is intended to quantify and measure the key performance activities in the game of Survivor. This site consists of four main sections, each section having two subsets of content:
Out Wit: consisting of Vote Accuracy and Vote Clustering
Out Play: consisting of Immunity Idols and Challenge Performance
Out Last: consisting of Vote Avoidance and Safety Earned
Scores: consisting of Game Scores and Player Scores
Additionally, the site has a reference page for all 46 seasons so that visitors can see season specific information. There is also a reference page for players, which allows visitors to find specific players and get to their respective player pages (by clicking on the player picture). Lastly, there is a tribal roster page so visitors can see the tribal breakdown by season.
Performance Measurements:
These particular set of measurements were selected because they represent key activities in the game, but they also qualify under three important criteria. In order to be considered, a game performance activity must be observable, measurable, and objective.
Observable:
For example, an observable activity is voting. At the end of each episode, every person's vote is displayed, so the activity can be tracked because it can be observed. Strategy on the other hand is much more difficult to consistently observe. Sometimes the viewers are shown players articulating or displaying strategic acumen, but it's not always shown. Or it's not shown for all players. Assigning credit to strategic initiatives in this sense is rather difficult, because the viewers are only shown what fits the story narrative, not necessarily what is happening from every player's viewpoint.
Measurable:
Challenge wins can be measured. Every challenge has a definitive winner and by default, all players who don't win are losers (of the challenge). But recording challenge performance beyond first place is difficult because it's not always clear who finishes in second, third, fourth, etc. For example, when a tribe (or individuals) are completing a puzzle at the end of a challenge, how is it to be determined who is in second place when the first place finisher is done? Usually, once the challenge is won, the competition is over. Is second place determined by percentage of puzzle complete? Is it determined by the most number of pieces placed? Even when a puzzle is mostly complete, pieces might not be placed correctly and that might only be known at the very end of the puzzle. But determining how long it might take to finish can't be done objectively. Additionally, even if the absolute placement was determined in a challenge (i.e. it played out until each tribe or individual completed the challenge), the final placement for a challenge is only ordinal. It's not relative. Meaning, it doesn't reflect how close second place was to winning. Does the tribe or individual who finishes second lose by seconds or minutes? For the winner, it doesn't matter if they win by seconds or minutes, because they completed the challenge first, which was what qualified them to win. But the value in measuring second place (and beyond) is to value how closely they came to winning. Since that can't be measured (or properly valued), only challenge wins are measured.
Objective:
Performance measurements must also be objective, not subjective. Who a player voted for is objective. It is clear, factual, and cannot be disputed. While subjective analysis is important and has it's place in the discussion of player performance, that is not the intention of this measurement system. The examples that follow provide insight as to why subjective analysis can be problematic from a measurement standpoint.
The first subjective challenge is trying to interpret strategic intent from how players vote (as in who they choose to vote for). Two examples demonstrate the difficulty in subjective analysis. The first example is from season 35 (cycle 14). Ben is considered to be public target number one after two failed attempts to vote him out. Chrissy, Ryan, Devon, and Mike agree that they have to win immunity and vote Ben out. What they don't know is that Ben secretly found a third idol during the night. Ben states during a confessional that his choice on whom to vote for is between Devon and Mike (as Chrissy is immune and Ryan is not competitive in challenges). Devon suspects that Ben might be acting, as he appears to be giving up. He tells Ryan that he is considering voting for Mike as a failsafe in case Ben plays an idol. Devon's concern is that he might anger Mike if he votes for him and Ben doesn't have an idol. After the votes are cast, Ben does in fact play his idol (cancelling the three votes against him), so only Ben's vote for Devon and Devon's vote for Mike stand. On the re-vote, Devon is saved and Mike is eliminated. Devon's instinct (and subsequent vote for Mike) ultimately save him from elimination. This is an example where voting incorrectly, is done intentionally to protect himself in case of an idol play.
The second example is from season 46 (cycle 12). While at the reward meal, Maria, Q, and Ben agree to vote out Venus. Meanwhile at camp, Kenzie, Venus, and Liz want to vote out Q. Later, both groups vie for Charlie's swing vote. Maria tries to convince Liz to vote Venus out. After a sketchy interaction with Venus, Charlie leans towards voting Venus out. He subsequently confers with Ben, who argues strongly that Venus is the best person to vote out. Charlie then speaks with Kenzie and makes the argument as to why they should vote Venus out. While the show doesn't definitively show us that Charlie, Liz, and Kenzie have all decided to vote out Venus (to maintain dramatic tension), we know after the vote that the only person left out of the vote is Venus. Everyone votes to eliminate Venus except for Venus (who votes for Q) and Ben (who votes for Kenzie). Despite Ben's stated intention to vote for Venus, he votes to eliminate his close friend and ally Kenzie. From a strategic perspective, it could be argued that he voted for Kenzie in case Venus played either an idol (which she indicated to Charlie that she had) or her shot-in-the-dark (both of which could cancel out votes cast against her). In this scenario, Kenzie could be eliminated, therefore keeping Ben safe. Fortunately, on the next episode we see Ben admit that he made a mistake in voting for Kenzie. He said he was tired and his mind went blank, therefore he wrote down the only name he could remember. In his confessional (where he has no incentive to be dishonest), he completely dispels any notion that his vote was strategic. But without this confessional, one might assume that his vote was strategic.
Therein lies the inherent problem with subjective analysis when looking at intentionality of voting. We see in these two votes that sometimes it's in a player's best interest to not vote for the person that they believe is being eliminated (Devon's vote). And sometimes player's just make bad decisions, not based on strategy, but based on the physical conditions and demands of the game (Ben's vote). So rather than try to discern strategic intent, the vote accuracy measurement is purely objective and based on the premise that voting correctly is generally a positive outcome.
Game Score Correlation to Final Placement:
The simplest way to rank player's performance is by observing the order in which they finished the game. As the goal of the game is to progress to the end, logic dictates that the closer a player finishes towards the end, the better they have performed. This implies that a 3rd place finisher has outperformed a player who placed 4th. But the history of this game is littered with examples where this doesn't make subjective sense. One such example was in season 39. Lauren finished in 4th place, but was the odds on favorite to win the game had she been able to reach final tribal council. This was precisely why Noura chose to put Lauren in the fire making challenge against Dean. Noura on the other hand, finished in 3rd place and received no votes to win the game. Does Noura finishing one spot ahead of Lauren prove that she outperformed Lauren? The other issue with using final placement, is that there is no ability to compare players from different seasons. Did Adam Klein (season 33 winner) outperform Sarah Lacina (season 34 winner)? There's no objective way to discern this unless a measurement system is developed that analyzes more than just final placement.
But in order to quantify the value of an overall performance measurement system, the system must be tied to some objective ranking. While final placement isn't a perfect measurement of performance, there is logic in using it as a reference point. The physics of the game dictate that a player must reach the final tribal council in order to win the game, but must also do so in a way that the majority of the jury (a player's eliminated competition) will vote for them to win, relative to their other choices. The meticulous and delicate balance between advancing in the game safely (avoiding elimination) and getting to the end with enough win equity (having played the game in a meaningful way) is a constant struggle for players. It's a balance between risk/reward, aggression/passiveness, knowing when to make a move and knowing when to hold tight. It's important to understand that players can fail either way, therefore it's not absolute to state that a player who finishes 3rd is automatically the player who outperformed a player who finished 4th.
The performance measurement system of this site is intended to simultaneously give more analysis and nuance to a player's actual performance (based on key activities in the game), but also highly correlate to the final placement. The final correlation calculation (performance to final placement) can be found in Game Scores. This allows analysis of player's performance within season, but also the ability to compare player's performance across seasons.
Future Additions:
The initial idea for this site was to create performance metrics for players to quantify which players performed the best in the game of Survivor. With that stated, there is so much in the game that can be analyzed using a data driven approach. Future topics could include the following (found under the Musings section):
Vote Entropy
Relative Value of the Idol
Measurement of Strategic Relationships
Elimination Velocity
Decision Making through Win Probabilities
Player Performance by Player Codex
Thank you so much for visiting and please feel free to share your input or feedback about any aspect of the site. Stay up to date by Subscribing (at the bottom of the homepage).
Comentarios